Elizabeth,
You could make a decent case for me being about 1/2 gene away from washing my hands 30 times a day.
Heat DOES kill electronics, no question about it. The devices most prone to heat effects are power devices, which obviously use plenty of heat sinks. ICs can cook, too, some of which have extremely high circuit density. The proof can be found in any semiconductors 'reliability' testing program where devices are tested to failure.
For example, modern multi-core CPUs will dissipate maybe 70 watts? Maybe more...maybe less, I'm not current. I know voltage requirements for some devices has dropped to keep power down. And look at the obsessive lengths some computer modders go to ensure proper cooling.
I'm waiting for a passively cooled class 'a' amp with heatpipes or maybe liquid cooling, chiller and pump.
That being said, such shorter lifespan for hotter stuff has a statistical base. Silicon based semiconductors simply don't like temps much above....say 150c which depending on how much power you're talking about may actually kick out quite a bit of heat. Example:: A Penny at 150c has a lot less heat energy than say......an anvil at room temp. A power transistor running hot in a properly designed situation....proper thermal contact and enough heatsink area and mass, will get the heatsink pretty warm.
The observation you may want to make is how hot is the EQUIPMENT in your 80f room? If the gear is in an enclosed space with poor or marginal ventilation, your 'goose' is cooked and you may just be lucky. OTOH, if your stuff is in a well ventilated space and is the good gear I know you like, than you'll be fine. Even Bryston can be cooked. They design stuff with the 'noise' of actual use in mind. If EVERYONE used the amp in a cool, well ventilated space, they wouldn't use as much heatsink. But, they were thinking ahead. You are in the normal, expected range of users. |
The amount of heat your gear will 'give up' to the environment is related to the difference in temp. If your amp runs at 100f in a 100f room, very little heat will be transferred. In a 65f room, you've got no problems.
During the hot part of the year, at least make sure you are well ventilated or perhaps even install a fan near any hot gear.....But be careful not to just blow at hot stuff, if you restrict the natural convection, you could possibly even make something run warmer...while trying to cool it. Any fan I place is above the hot piece and blowing AWAY to suck air thru the piece and allow 'nature' to do its work.
And from our FWIW department, those nutty radio astronomers run their amplifiers and much electronics in a liquid nitrogen bath at colder than -320f.
This is to reduce the 'thermal' noise of atoms bumping into one another. In the very highest resolution systems this may make a difference. Or not. LOL.
Other temp effects may be on the speakers where characteristics of some synthetics....perhaps woof cones, may change from the coldest to warmest temps encountered. I'd also suspect greater power handling capacity in the coldest weather. The voice coils will cool much better. |
I see how I miswrote. But I'll stand by.....that as the temperatures of the room and amp get closer together, heat transfer slows.
Point is, cooler is better and you can cook it in a hot environment.
Heat and temperature are 2 different things.
And NO, the amp won't always be the same temp delta from ambient. In SS, for example, you have a max temp possible....say the junction temp of the devices. In a hot room wont' the difference drop as the room temp approaches junction temp? Or will the junction keep getting hotter until failure? Isn't there an upper limit to the temp of an amp?
2 amps of identical efficiency and power rating being run identically will be at different temps depending on the mass of the amp. And we all know how much heatsinks cost.
It's kind of an aside, but look at a few Stereophile amp tests where they 'precondition' an amp at 1/3 power for an hour before bench measurements. Some amps fail.
I'll call a friend of mine who is a PHd in physics. he'll straighten me out. His area of expertise is semiconductors, so it'll be good info. |
Let's straighten out the question? OK?
The OP wanted to know if heat was OK, and if the huge seasonal temp fluctuation was OK....
Well, I think we all agree that heat in excess is bad for electronics. Cold, especially condensing cold is perhaps worse....ZAP.
Heat cycling can also damage gear. Can Expansion and contraction of solder connections work them loose? Some of the new solders are less malleable than in years past. Wasn't that one of the problems with the X-Box?
Do we agree that cold air is better at sinking heat from electronics? It would seem that as the ambient temperature and temperature of the electronics got closer and closer, the amount of HEAT transferred would get less and less. It maybe that BigBucks is right, but I don't see it. The constant delta above ambient may work but I just see stuff getting hotter faster than the room it's in....especially if the room is externally heated...sunlight, hot day...etc. At some point, the junction temp of an output device would be nearing limits and be unable to dump enough heat.....thru all forms of shedding...radiation, conduction, convection....(others?) But would that be at a constant delta from ambient?
The electronics would raise temperature as the amount of heat soaked away got less but that would catch up to you at some catastrophically high temp....which would be a much higher temp that you'd like your room!
No matter the physics equities here, I still think that a room of 85f is WAY too hot for good electronics. Maybe just sitting there....OK, but I'd never run my TV in that hot a space. Or even my 'd' amp.
After living in the same house for 20+ years, I installed AC before last summer. Glad I did, too.
And Al, I agree with you, too. Starting from 'cold' stuff starts shedding heat as stuff warms. Convection. Conduction. Radiation. All play a part in shedding heat. However, that heat goes somewhere. A bad / Extreme example is my RPTV. It kicks out a jumbo amount of heat. That lamp COOKS. Well, it sort of keeps the house thermostat artificially WARM. The TV is about 6' from the thermostat. The rest of the house cools and gets downright cold.... But that TV warmed thermostat says that all is well.
I don't mean to play the 'expert' card, but I will call my physics buddy. He is a hi-end semiconductor engineer and should be conversant with these issues. I'll ask and post back..Give me a couple days. If I have to buy him lunch, I'm billing you guys for 1/3 of the bill....each! just kidding. |
Al, computer cooling, certainly a 'side issue' here is related to hifi. The point being that 'heat kills'. Enough statistical data exists to support this.
I had a Motherboard which had an automatic overclock feature when the CPU was pushed. I kept it on the most conservative setting and ran one of the Zalman Cu/Al 'mushroom' shaped heat sinks with the fan on 'full'. I chose a conservative CPU and never pushed it. I never saw my 2.4gig CPU go above about 2.6 I kept the dust bunnies cleaned and the all the cooler fins de-linted.
Point is valid that 'heat kills'.
Processing temperatures for Silicon devices run from about 1150c for some of the Junction Drives down to very low temps like 200c or less, used for 'sinter' or 'anneal' processes...usually right at, or near the 'end of the line'. Nothing goes above about 425c after 'metalization' which is usually an aluminum alloy and deposited somewhere mid-line. I am not current on what is used in State of the Art CPUs. They may have gone to copper or some other metal. Very thing / narrow aluminum has some problems with reliability and electromigration.
The 'statistics' to which Paperw8 also refers to are valid. They are well understood and proven. That they are statistics means you are dealing with populations.....large numbers of a given object some of which will crap out immediately and others which will last.....seemingly forever. The 'take away' is that Al is also correct: MTBF is the most visible metric applied to this stuff. The object of manufacturing quality is to produce product in the middle of the spec. All 'excursions' are suspect.
Semiconductor plants, called 'fabs' spend a bundle on 'rel labs'....Reliability. Here they torture what they make in an 'accelerated' lifetime. Indeed, before a new product or process is released to production, parts must go thru what is called a '1000 hour burn-in'. Too many failures or parametric shifts during the test are grounds to deny the 'go ahead' for volume production. What went wrong? The innocent are usually than punished.
I worked at a company that had a 'Hi/Lo' group. Bad lots were investigated....what went wrong? Especially GOOD lots got the same treatment....what went right? Go Figger.
Now, if you go for the 'weakest link' line of thought, buying mil-spec stuff for part of your design and cheap-o commodity chips for other parts doesn't make sense. This is why good equipment is not only well designed...but well executed, too. One could probably build a Bryston Copy for a fraction the cost of the real thing but out of cheaper parts and NO warranty! It wouldn't surprise me to learn that companies like Bryston had 'thermal budget' folks on hand. Engineers to calculate and verify type and amount of heat sinking. Even those IR imagers to look for 'hot spots'. All sorts of cool, hi-tech lab gear.
Mil Spec parts are not necessarily different from those you can buy. I say necessarily because electrically they may be the same and even made in the same fab. The difference is that NO rework is allowed, so if a production 'lot' has a problem, it is immediately either scrapped or degraded to 'consumer'. Other rules apply and the factory audits are BRUTAL. A factory must EARN the right to build Mil-Spec parts and be so certified. Than periodically audited. The taxpayer picks up the bill. You'd use parts like that, too, if a repair call took you 200 miles into space. |
http://www.ee.latrobe.edu.au/internal/workshop/store/pdf/MJ2955.pdf
First datasheet I came across.
Look at figure #1 for temp derate and the bottom of page #1, just above the fig for the heat/ thermal #s. |
This is baffling. But I DO see what you mean. Since a device and its enclosure/ sinking can move only so much heat.....so fast.... the junction will be above ambient as its heat migrates away. And I see they should stay a set difference apart. As long as you can sink the whole system.
But what happens in an enclosed space? I've had gear in confined spaces where the heat evolved simply had no place to go. Would the temp difference continue until the device failed? I'm talking Very Hot.....like over 100c, air temp. perhaps.
What happens if you put a cold / off device into a warm environment? Does the device warm....than as its temp rises to the 'delta' temp, heat begins moving the 'right' way?
My 'd' amp is on the shelf below my small dish receiver. If I close the door overnight, even with everything off/ standby, the next morning it is pretty warm inside. Even the amp is warm to the touch.
I walked into a very small demo room at a video store. They had 4 plasma TVs in about a 10x10 foot room......and it was almost too hot to breath. I'm sure the electronics was way too hot for comfort.
I spent 25 years building semiconductors from wafers to die. Apparently I didn't spend enough time in probe or reliability.
|
HiFi,
Let me give this a try.....Big, check and see if we're getting this.
The semiconductor will, under use simply go to some temp above ambient and stay there, depending on load. The device HAS to be warmer than the heatsink.
I don't know...If you chill the sink to 30f while the device is at 90f? The device will first cool, than begin pumping heat back into the system raising its temp.
HERE:: READ THIS.....
I'm going to go thru it. I already see it makes sense of this and the math part isn't too bad.
http://homepages.which.net/~paul.hills/Heatsinks/HeatsinksBody.html
BigBucks. Please read above link and see if it meets your approval. It seems straightforward explained this way.
One further question, however.......and that deals with maximum power dissipation of semiconductor devices as well as the maximum junction temp allowed.....all apparently related back to ambient temp and the thermal resistance of the system.....
cheers........ |
Hifi,
You can always run something out of spec and blow it up.
Read the link I provided.
I have no idea of the speed of propagation from the junction, thru all the various interfaces and to the air.
Maybe you COULD 'shock' it.
C or F? just a conversion. The US is probably the ONLY country left in the world who uses the English metrology system. Darn French. |
HiFi.
Big is correct. Devices are apparently designed with a max junction temp at a given air temp. Lower air temp is gravey.
Read at least the written description of the math. It'll make sense. The heat-engine model doesn't apply where greater differences in temp produce greater results. transistor temps are sort of self-limiting and there are bunches of design criteria.
That being said, cooler is still better. |
As near as I can figure, that's the whole point. You'll have a zero difference when you fire it up with everything at room temp....BUT as soon as the device warms, it'll start moving heat away at such a rate as to maintain the calculated (or nearly) temp diff above ambient.
If you stick your room temp amp into the freezer, same deal.....just that now your working against the freezers ability to pull heat out (pump it) 'uphill' into the room....creating the cold box. Amp doesn't know this and will still end up warm VS the internal cold box temp.....and the same # of degrees.
If you put a power transistor in an insulated space and power it up,
you'll have a meltdown / failure in no time. No place for the heat to go. somewhere above maybe 150 to 200c, it'd just cease to function. These limits are all straight physics and chemistry.
But, based on the math the device will shed heat at a certain rate. Once it's been on and is stable, it'll run pretty much a certain amount above ambient for a long time. Even in the freezer......
This is pretty clear from the link I posted.
Also, find a datasheet. Somewhere is the 'derate' for power devices...and maybe others. The derate deals with power and temp. And how much less the device will take and at what rate, as it warms.
The OP? Long gone, but still and all, I'd not run my gear in an 85f space. Any weak link will be ruthlessly exposed. Any dry heatsink compound which isn't doing its job....a nut/screw securing a power transistor has come loose. A dust bunny clogging heatsink fins or some venting,.... All can hurt your stuff. |
Lesson Learned:
In Italy their is a very nice Roman Aqueduct bridging a river. It is 3 layers high and is in perfect shape after who knows how long? If needed, it'd probably be easy to put it back in service.
Recent analysis shows it to be built to 'modern' standards of about 2:1 over the maximum anticipated stress.\
Article was in Scientific American, so I could probably look it up if anyone was curious.
Running stuff at redline is a sure-fire start of problems. |
If I felt the need to design something needing heat management, I'd try to over engineer by at least a factor of 2x. This would be using all rule of thumb estimates and data sheet numbers. If a doubt existed, buy the next larger heat sink. More space between caps. Vent the transformer. Any wacky thing I could think of to shed heat.
If I could work the math, I could probably cut it closer and save money, time and bulk.
Read up on some heat management issues. Just an example:: If the heat sink is fins are up / down the natural convection will funnel air up. You should provide venting above and below to facilitate this flow.
Put the SAME heatsink horizontal, with the same load, and suddenly you have way too little heat sink.
Add some forced air to EITHER and you are ahead. Forced air fans may require or tolerate different fin spacing.
I look at the ratty heatsink from an old CPU. What an awful design. Fan blew down into it and then out the sides. Problem? Well, I can't imagine much airflow in the center, at the base of the fins. I use this otherwise worthless extrusion as a letter holder. Works GREAT for that!
Point? BigBucks is right. There IS a substantial science behind this stuff. Wouldn't surprise me to hear of very close heat budgets in aerospace applications where every ounce shot into space requires gallons of fuel and every cubic inch counts. Knowing EXACTLY how much heat at what junction temp and how fast it migrates is within the realm of 'knowable'..... |
